Sulfatase 1 Promotes the Motor Neuron-To-Oligodendrocyte Fate Switch by Activating Shh Signaling in Olig2 Progenitors of the Embryonic Ventral Spinal Cord

18018 • The Journal of Neuroscience, December 12, 2012 • 32(50):18018–18034

Development/Plasticity/Repair Sulfatase 1 Promotes the Motor Neuron-to-Oligodendrocyte Fate Switch by Activating Shh Signaling in Olig2 Progenitors of the Embryonic Ventral Spinal Cord

Yacine Touahri, Nathalie Escalas, Bertrand Benazeraf, Philippe Cochard, Cathy Danesin, and Cathy Soula University of Toulouse, Center of Developmental Biology, F-31062 Toulouse, France

In the developing ventral spinal cord, motor neurons (MNs) and oligodendrocyte precursor cells (OPCs) are sequentially generated from acommonpoolofneuralprogenitorsincludedintheso-calledpMNdomaincharacterizedbyOlig2expression.Here,weestablishthatthe secretedSulfatase1(Sulf1)isamajorcomponentofthemechanismthatcausestheseprogenitorstostopproducingMNsandchangetheir fate to generate OPCs. We show that specification of OPCs is severely affected in sulf1-deficient mouse embryos. This defect does not rely on abnormal patterning of the spinal cord or failure in maintenance of pMN progenitors at the onset of OPC specification. Instead, the efficiency of OPC induction is reduced, only few Olig2 progenitors are recruited to generate OPCs, meanwhile they continue to produce MNs beyond the normal timing of the neuroglial switch. Using the chicken embryo, we show that Sulf1 activity is required precisely at the stage of the MN-to-OPC fate switch. Finally, we bring arguments supporting the view that Sulf1 controls the level of Sonic Hedgehog (Shh) signaling activity, behaving as an enhancer rather than an obligatory component in the Shh pathway. Our study provides additional insights into the temporal control of Olig2 progenitor cell fate change by the identification of Sulf1 as an extracellular timing signal in the ventral spinal cord.

Introduction gions (Briscoe and Novitch, 2008). Among them, the pMN Neurons and glial cells of the CNS originate from neural progen- domain, marked by expression of Olig2, first generates all of the itors that differentiate in a highly specific manner, both spatially motor neurons (MNs) and changes with time to produce most and temporally. The spinal cord has served as an excellent model oligodendrocyte precursor cells (OPCs) of the spinal cord (Row- for studying both diversity and successive waves of neural cell itch, 2004). How this developmental switch from production of generation. Under the influence of patterning signals, the neural MNs to OPCs is controlled has been partially elucidated, and tube is segmented into distinct progenitor domains, each dedi- both cell-intrinsic and cell-extrinsic mechanisms were shown to cated to generate a particular neuronal and glial cell population. underlie this coordination (Kessaris et al., 2008; Rowitch and In the ventral neural tube, depending on both its concentration Kriegstein, 2010). Strikingly, signaling factors that control OPC and timing of exposure, Sonic Hedgehog (Shh) regulates expres- generation in the spinal cord have much in common with those sion of transcription factors that demarcate five progenitor re- that instruct neural progenitors to become specialized neurons. In the ventral spinal cord, the decision to produce OPCs, at a time long after the dorsoventral neuronal patterning is completed, still Received July 25, 2012; revised Oct. 2, 2012; accepted Oct. 21, 2012. depends on Shh activity (Poncet et al., 1996; Pringle et al., 1996; Author contributions: C.S. designed research; Y.T., N.E., B.B., and C.D. performed research; Y.T., N.E., B.B., P.C., Orentas et al., 1999; Soula et al., 2001; Agius et al., 2004; Park et C.D., and C.S. analyzed data; C.S. wrote the paper. ThisworkinthelaboratoryofC.S.wassupportedbygrantsfromtheNationalAgencyforResearch,Associationfor al., 2004). In chicken, the specificity in the response to Shh relies Research on Multiple Sclerosis (ARSEP), National Center of Scientific Research, and University of Toulouse. Y.T. was on its level of activity, optimal OPC induction being driven at supported by doctoral grants from Ministry of National Education, Research, and Technology and ARSEP. We ac- higher doses of Shh than those required to generate MNs (Soula knowledge the Developmental Studies Hybridoma Bank, developed under the auspice of National Institute of Child et al., 2001; Danesin et al., 2006). Accordingly, a temporal varia- HealthandHumanDevelopmentandmaintainedbytheUniversityofIowa,DepartmentofBiologicalSciences(Iowa City,IA),forsupplyingmonoclonalantibodies.WearegratefultocolleaguesintheCenterforDevelopmentalBiology tion in Shh distribution, concentrating at the apical surface of for help, encouragement, and advice, especially B. Glise and F. Pituello for helpful discussions and critical reading of ventral neural progenitors, has been proposed to trigger this cell this manuscript. We especially thank X. Ai and C. Emerson for their generous sharing of the sulf1 mutant line and R. fate change (Danesin et al., 2006). Other signaling molecules, Zeller for kindly sharing shh mutant line. We thank Yannick Carrier and Audrey Montigny for invaluable technical such as FGFs, have also been recognized as OPC inducers, in assistance and Jacques Auriol for assistance with mouse breeding and care. We are grateful to D. Anderson, F. Guillemot,A.Joyner,M.Scott,C.Tabin,andA.Vincentwhokindlyprovideduswithvaluablereagents.Wethankthe particular in the dorsal spinal cord, but do not seem to be active in Toulouse Regional Imaging Platform for technical assistance in confocal microscopy. its ventral counterpart (Chandran et al., 2003; Kessaris et al., The authors declare no competing financial interests. 2004). Finally, the dorsalizing factors, BMPs and Wnt, by repress- Correspondence should be addressed to Cathy Soula, University of Toulouse, Center for Developmental Biology, ing the OPC fate, are also regulators of the decision to make Coeducational Research Unit 5547, National Center of Scientific Research/University Paul Sabatier, 118 Route de Narbonne, F-31062 Toulouse, France. E-mail: [email protected]. neurons or OPCs (Mekki-Dauriac et al., 2002; Miller et al., 2004; DOI:10.1523/JNEUROSCI.3553-12.2012 Bilican et al., 2008; Langseth et al., 2010). How these signaling Copyright © 2012 the authors 0270-6474/12/3218018-17$15.00/0 pathways are temporally controlled to confer distinct identities to Touahri et al. • Sulf1, An Activator of Shh-Dependent OPC Fate Choice J. Neurosci., December 12, 2012 • 32(50):18018–18034 • 18019 neural progenitors remains an opened question. Sulfatase 1 following monoclonal antibodies (culture supernatants) were obtained (Sulf1), a regulator of the heparan sulfate proteoglycan (HSPG) from Developmental Studies Hybridoma Bank: anti-MNR2 at 1:8 sulfation “code” (Dhoot et al., 2001; Morimoto-Tomita et al., (Tanabe et al., 1998), anti-Nkx2.2 at 1:2 (Ericson et al., 1996), and anti- 2002; Ai et al., 2006; Lamanna et al., 2006), represents a good Islet1 at 1:3 (Ericson et al., 1992). Alexa Fluor 488, 555, or 647 goat candidate to orchestrate such temporal variations (Braquart- anti-rabbit, goat anti-mouse, donkey anti-goat, or donkey anti-rat at 1:500 were from Invitrogen. Rabbit anti-Sulf1 antibodies, directed Varnier et al., 2004; Danesin et al., 2006). HSPGs are crucial against the hydrophilic domain of chicken and mouse Sulf1, were gener- players in numerous signaling pathways, and Sulf1, by remodel- ated as described previously and used diluted at 1:200 for immunodetec- ing their 6O-sulfation pattern, is recognized as a key regulator of tion experiments (Braquart-Varnier et al., 2004; Ai et al., 2007). their interactions with signaling molecules, including Hedgehog In situ hybridization. In situ hybridization (ISH) was performed on proteins (Danesin et al., 2006; Ratzka et al., 2008; Wojcinski et al., vibratome sections of mouse and chicken embryos and spinal cord ex- 2011). A possible relationship between Sulf1 activity and the MN- plants either by hand or automatically (InsituPro, formerly Intavis) using to-OPC switch was initially proposed based on the tight correla- the whole-mount ISH protocol described previously (Braquart-Varnier tion between Sulf1 dynamic spatiotemporal expression pattern et al., 2004; Danesin et al., 2006). Digoxigenin (DIG)-labeled sense and and the timing of Shh-dependent OPC induction in the chicken antisense RNA probes were synthesized using T3 and T7 polymerases. ventral spinal cord (Braquart-Varnier et al., 2004; Danesin et al., RNA labeled probes were detected by an alkaline-phosphatase-coupled antibody (Roche Diagnostics), and nitroblue tetrazolium/5-bromo-4- 2006). However, direct evidences for Sulf1 function in regulating chloro-3-indolyl phosphate (NBT/BCIP) was used as a chromogenic OPC generation were still lacking. substrate for alkaline phosphatase (Boehringer Mannheim). Sections In this study, we show that Sulf1 activity is required for proper were processed for two-color double hybridization using DIG- and specification of OPCs in the embryonic ventral spinal cord. Our fluorescein-labeled antisense riboprobes for sulf1 and either patched1 results support a model whereby Sulf1 acts as a temporal regula- (ptc1)orgli1. DIG-labeled probes were revealed in dark blue using NBT/ tor of Shh activity that triggers Olig2 progenitor cell fate change, BCIP solution, whereas fluorescein-labeled probes were revealed in red from the production of MNs to that of OPCs. using INT/BCIP solution, following standard procedure. Stained sections were digitized on a Nikon microscope, and images were manipu- Materials and Methods lated using Adobe Photoshop (Adobe Systems). The following gene probes were used: chick and mouse ptc1 [provided by C. Tabin (Harvard Animals Medical School, Boston, MA) and M. Scott (Standford University School Experiments were performed in accordance with the European Commu- of Medicine, Standford, CA), respectively], chick and mouse gli1 (pro- nity guiding principles on the care and use of animals (86/609/CEE; vided by C. Tabin and A. Joyner, Sloan-Kettering Institute, New York, Official Journal of the European Communities number L358, December NY, respectively), mouse ebf2 (a gift from A. Vincent, Center fo Devel- 18, 1986), French decree number 97/748 of October 19, 1987 (Official opmental Biology, Toulouse, France), mouse ngn2 (provided by F. Guil- Journal of the French Republic, October 20, 1987), and the recommen- lemot, National Institute for Medical Research, Mill Hill, London, UK), dations of the National Center of Scientific Research. Generation of Shh sulf1 (provided by X. Ai, Boston University School of Medicine, Boston, and Sulf1 mutant mouse lines were reported previously (St-Jacques et al., MA), and plp/dm20 (a gift from J. L. Thomas, Paris, France). Counter- 1998; Ai et al., 2007). Mutant mice were maintained as heterozygous staining of Olig2 was performed after color development after a postfix- stocks in C57BL/6 genetic background and crossed to generate different Ϫ Ϫ ation step in 4% PFA for 1 h. mutant combinations. For genetic interaction experiments, sulf1 / and shhϩ/Ϫ mice were mated to generate sulf1ϩ/Ϫ;shhϩ/Ϫ heterozygous mice, which were further crossed with sulf1Ϫ/Ϫ mice to generate sulf1Ϫ/Ϫ; Isolation and culture of neural tissues ϩ Ϫ Ϫ Ϫ ϩ ϩ Flat-mount preparations of spinal cord explants were cultivated using an shh / and sulf1 / ;shh / littermate embryos. Genotyping was per- organotypic culture system described previously (Agius et al., 2004). formed by DNA extraction from tails followed by PCR (St-Jacques et al., Briefly, spinal cords were manually dissected from chicken embryos at 1998; Ai et al., 2007). Embryonic day 0.5 (E0.5) was the day vaginal plugs E4.5. Spinal cords were isolated from surrounding tissues and opened were found. Mutant and control mice of either sex were analyzed from along the dorsal midline. The cervico-brachial region was isolated, and the same litters, facilitating comparisons among the genotypes. The this explant was flattened on a nitrocellulose membrane (Sartorius) with stages of embryos with different genotypes were further matched using neural progenitors up and further grown in DMEM (Invitrogen) supple- size and shape of limb buds. Fertilized White Leghorn chicken eggs, mented with 10% FCS (Sigma). For Sulf1-neutralizing experiments, the obtained from a commercial source, were incubated at 38°C until they antiserum (␣Sulf1) was added to culture medium at 1:100 and 1:50 di- reached the appropriate stages (Hamburger and Hamilton, 1992). lutions, and preimmune serum was used as control (Gill et al., 2010). For Staining procedures BrdU staining, spinal cord explants were incubated for 2 h with BrdU Tissue preparation. Mouse and chicken embryos or flat-mounted ex- (0.15 ␮g/␮l; Sigma-Aldrich), 24, 32, or 48 h after plating. Tissues were plants were fixed in 4% paraformaldehyde (PFA) in PBS overnight at then fixed for 4 h and processed for BrdU immunostaining. 4°C, dehydrated, and stored in 100% ethanol at Ϫ20°C. Tissues were then sectioned at 60–80 ␮m using a vibratome (Microm). In all experi- Electroporation ments, sections were performed at the cervico-brachial level. For detec- E4.5 spinal cord electroporation was performed ex ovo. Chicken embryos tion of Sulf1 on spinal cord or explant sections, we used the previously were harvested and isolated in a Petri dish with the dorsal side up, and reported protocol (Gritli-Linde et al., 2001; Braquart-Varnier et al., DNA solution was injected into the lumen of the spinal cord as described 2004) that allows preservation of extracellular matrix molecules. Speci- previously (Danesin et al., 2006). Electrodes were positioned on each side mens were fixed overnight in cold 95% ethanol and 1% acetic acid of the cervico-brachial region of the spinal cord, the positive electrode (Sainte Marie’s solution) and then washed in 95% ethanol. being placed more ventrally than the negative one, allowing satisfactory Immunostaining. For immunofluorescence analysis, chicken and electroporation of ventral regions. Ten pulses of 25 V were applied, and mouse tissues were processed as described previously (Agius et al., 2004; spinal cord was further dissected and grown in organotypic culture as Danesin et al., 2006), except that horse serum (10%) was used to block above. All expression vectors were used at 1 ␮g/␮l. Expression constructs mouse spinal cord sections. Primary antibodies were applied overnight at were cloned into either the pCIG vector (a gift from A. McMahon, Har- 4°C. Secondary antibodies were applied at room temperature for 1 h. The vard University, Cambridge, MA) for quail Sulf1 (Ai et al., 2003; Danesin antibodies used were as follows: rabbit anti-Olig2 at 1:500 (Millipore et al., 2006) or pRFPRNAiA vector (Das et al., 2006) for sulf1RNAi. The Bioscience Research Reagents); goat anti-Sox10 at 1:500 and anti-Nkx2.2 sulf1RNAi vector was constructed according to Das et al. (2006) using a at 1:250 (Santa Cruz Biotechnology); rat anti-BrdU at 1:1000 (AbD Se- target sequence of 19 nucleotides (5Ј-AAAACAGAACAGAGUCUUAGAGG- rotec); and rabbit anti-GFP at 1:500 (Torrey Pines Scientific). All the 3Ј) designed on chicken sulf1 mRNA using the MWG RNAi Design Tool 18020 • J. Neurosci., December 12, 2012 • 32(50):18018–18034 Touahri et al. • Sulf1, An Activator of Shh-Dependent OPC Fate Choice

(MWG Biotech). In mutated form of Sulf1 (CC89,90AA), cysteines 89 OPC number with an average of 3.02 Ϯ 0.89 cells per hemisection and 90 have been replaced by alanines to block the N-formylglycine in sulf1Ϫ/Ϫ spinal cords (n ϭ 5) compared with 12.48 Ϯ 1.43 in modification, required for catalytic activity but not for substrate binding wild-type siblings (n ϭ 9; Fig. 1H). Such a decrease, although less (Dhoot et al., 2001). Empty pCIG and pRFPRNAiGFP (Das et al., 2006) pronounced than in homozygous mutants, was also observed in vectors were used as controls. sulf1 heterozygous embryos (6.66 Ϯ 1.28 cells, n ϭ 6; Fig. 1H and Imaging, cell counting, and statistical analyses data not shown), suggesting a dosage-dependent regulation of Vibratome tissue slices were analyzed with Leica SP2 and SP5 confocal ventral OPC development by sulf1. We did not observe differ- microscopes. For analyses of mouse tissues, identical confocal acquisi- ences in size and shape of the embryonic spinal cord between tion parameters were applied to sibling wild-type and sulf1 mutant em- wild-type and sulf1Ϫ/Ϫ sibling embryos, arguing against the pos- bryos. Mouse embryonic tissue slices were harvested from the level of the sibility that the defect in OPC generation could simply reflect a anterior limb bud. In Olig2 and Sox10 immunostaining experiments, delay in development of the CNS in sulf1 mutant embryos. Im- positive cells were considered to belong to the mantle zone when their portantly, the number of Olig2-positive cells in the progenitor nuclei were clearly individualized from the Olig2-positive cells of the zone appeared similar between wild-type and sulf1 mutant em- progenitor zone (Fig. 1). Cell counting was performed on at least five tissue slices per embryo or chicken explant. For each tissue slice (60–80 bryos (Fig. 1F,G). Cell counting confirmed that the number of ␮m), at least four optical sections were acquired at 6 ␮m intervals, and Olig2-expressing cells in the progenitor zone was not affected in cells were counted in each optical section. The data provided are the E12.5 sulf1-deficient embryos (43 Ϯ 1.52 in sulf1 mutant, n ϭ 3, average of at least three embryos or explants (n) per condition from at vs 43.3 Ϯ 1.27 in wild-type embryos, n ϭ 4; Fig. 1H), indicating least two independent experiments. Quantifications are expressed as the that Sulf1 is dispensable for formation of the pMN domain mean Ϯ SEM number of cells in an optical section of an hemi-spinal cord and/or maintenance of Olig2 progenitors. However, in our ex- or of an hemi-explant. Statistical analyses were performed using the periments, we noticed differential intensities of the Olig2 signal Ͻ Ϫ Ϫ Mann–Whitney U test. Significance was determined at p 0.05. p values in the progenitor zone of E12.5 sulf1 / embryos that appeared are indicated in figure legends or in text when quantifications are not lowered compared with wild types (Fig. 1, compare F, G). included in figures. Ϫ Ϫ The reduction in the number of migrating OPCs in sulf1 / embryos was still apparent at least until E14.5/E15 (Fig. 1I,J). Results Quantification of Olig2-expressing cells in the mantle and mar- OPC development is impaired in sulf1-deficient mouse ginal zones revealed that the number of OPCs was reduced to embryonic ventral spinal cord 60% of the number counted in age-matched wild-type spinal In chicken, Sulf1, initially expressed in floor plate cells at stages of cords (125 Ϯ 20.9 cells in sulf1 mutant, n ϭ 4, vs 204.5 Ϯ 0.28 in neuronal generation, is further upregulated in the ventralmost wild-type embryos, n ϭ 4; Fig. 1M). The distribution of Olig2- neural progenitors of the spinal cord immediately before OPC expressing cells in sulf1Ϫ/Ϫ spinal cords, scattered throughout the specification, i.e., from E4.5/5 (Braquart-Varnier et al., 2004; dorsal and ventral mantle and marginal zones, was similar to that Danesin et al., 2006). We first asked whether a similar upregula- found in wild-type embryos, indicating absence of gross defects tion of sulf1 also occurs in mice and whether it correlates with the in migratory properties of OPCs generated in sulf1 mutant em- MN-to-OPC switch, known to occur between E11.5 and E12.5 bryos (Fig. 1I,J). Interestingly, in E14.5/E15 sulf1Ϫ/Ϫ spinal (Richardson et al., 2000). At E10.5, during MN generation, sulf1 cords, contrasting with data obtained at earlier embryonic stages, expression was restricted to floor plate cells (Fig. 1A). Its expres- we observed a reduction in the number of Olig2-expressing cells sion domain then extended dorsally, and, from E11.5, sulf1 within the progenitor zone (9 Ϯ 0.61 cells in sulf1 mutant, n ϭ 4, mRNA was also detected in ventral neural progenitors, adjacent vs 15.75 Ϯ 0.14 in wild-type embryos, n ϭ 4; Fig. 1K–M), indi- to the floor plate (Fig. 1B,C), indicating that the spatiotemporal cating that sulf1 loss of function led to a depletion in the pool of course of sulf1 expression defined in chicken is conserved in the Olig2 progenitors at these later stages. Expectedly, around birth, mouse. We then turned to a sulf1 knocked out (sulf1Ϫ/Ϫ) mouse it was no longer possible to distinguish wild-type and sulf1 mu- line to directly address the role of sulf1 in OPC development. tant spinal cords using Olig2 and proteolipid protein (PLP)/ Mice deficient for Sulf1 show no overt phenotype and are viable DM20 as oligodendroglial markers (Fig. 1N–Q), suggesting that and fertile, indicating that the enzyme is not essential for embry- dorsal progenitors and/or proliferation/survival of the fewer ven- onic development (Lamanna et al., 2006; Ai et al., 2007). Only trally generated OPCs might have contributed to replenish the subtle defects in neurite outgrowth have been reported in the pool of oligodendroglial cells. developing CNS (Kalus et al., 2009). Despite the lack of drastic Together, our results showing that the OPC number was phenotype, we postulated that this did not preclude defects in markedly reduced in E12.5 sulf1Ϫ/Ϫ embryos, a depletion still OPC development from ventral spinal cord progenitors. Indeed, apparent 2 d after the timing of the MN-to-OPC fate switch, in both brain and spinal cord, failure in OPC generation from a clearly point to a role for sulf1 in the control of ventral OPC single source can be compensated by overproduction of OPCs generation. from other sources (Cai et al., 2005; Vallstedt et al., 2005; Kessaris et al., 2006). We analyzed ventral OPC generation in sulf1- Sulf1 is required for efficient OPC specification from deficient embryos using Olig2, a reliable marker of OPCs within Olig2 progenitors the mantle and marginal zones of the developing spinal cord. We next asked whether the deficient OPC generation in sulf1 Indeed, in contrast to MNs, OPCs maintain Olig2 expression mutant embryos resulted from a failure in the initial commit- when they emigrate from the progenitor zone to reach their final ment of Olig2 progenitors toward the OPC fate. To address this destination (Lu et al., 2000; Takebayashi et al., 2000; Zhou et al., question, we used Sox10, the earliest known transcription factor 2000). At E12.5, soon after initiation of OPC generation, we de- expressed in oligodendrocyte lineage cells of the mouse spinal tected Olig2-positive cells dispersed in the mantle zone of wild- cord (Kuhlbrodt et al., 1998; Zhou et al., 2000; Stolt et al., 2002). type animals (Fig. 1D). In contrast, only few Olig2-positive cells Sox10 is specifically upregulated in Olig2 progenitors fated to were observed in the mantle zone of age-matched sulf1-deficient generate OPCs as soon as E11.5, and its expression is further embryos (Fig. 1E). Cell counting revealed a 75% reduction in maintained in OPCs migrating in the mantle zone (Zhou et al., Touahri et al. • Sulf1, An Activator of Shh-Dependent OPC Fate Choice J. Neurosci., December 12, 2012 • 32(50):18018–18034 • 18021

Figure 1. Sulf1 loss-of-function impairs OPC generation from Olig2 progenitors in the mouse embryonic spinal cord. A–C, Temporal expression profile of sulf1 on transverse sections of mouse embryonic brachial spinal cord at E10.5 (A), E11.5 (B), and E12.5 (C). Note the dorsal extension of the sulf1-expressing domain from E11.5 (arrows in B, C). D–O,(Figure legend continues.) 18022 • J. Neurosci., December 12, 2012 • 32(50):18018–18034 Touahri et al. • Sulf1, An Activator of Shh-Dependent OPC Fate Choice

2000). Confirming the phenotype observed using Olig2, the expression in cells adjacent to dorsal progenitors of the spinal number of Sox10-positive cells dispersed in the mantle zone at cord, whereas only faint expression was detected in cells facing E12.5 was reduced by 78% in sulf1Ϫ/Ϫ embryos (2.55 Ϯ 0.91, n ϭ the Olig2-expressing domain, in agreement with the fact that 5) compared with wild-type littermates (11.42 Ϯ 1.71, n ϭ 5; Fig. MNs were no longer produced at this stage (Fig. 3C). In contrast, 2F and data not shown). We then focused on Sox10 expression in in E12.5 sulf1 mutant embryos, a higher level of ebf2 expression the progenitor zone. At E11.5, there was a strong reduction in the was detected at the level of the pMN domain, indicating that MN number of Sox10-positive cells within the Olig2 progenitor do- genesis was still occurring at this late stage (Fig. 3D). Similarly, main of sulf1Ϫ/Ϫ embryos (0.56 Ϯ 0.16, n ϭ 4) compared with although the proneural gene ngn2 was not detected at the level of wild-type littermates (3.94 Ϯ 0.73, n ϭ 4; Fig. 2A–C). At E12.5, the pMN domain in wild-type embryos, cells emerging from this Sox10 expression was detected in almost 50% of Olig2 progeni- domain were detected in littermate sulf1 mutant embryos (Fig. tors in wild-type embryos (20.48 Ϯ 1.49, n ϭ 5), whereas Ͻ20% 3E,F), again supporting ongoing generation of neurons at a stage of them upregulated Sox10 in sulf1 mutant embryos (8.11 Ϯ 1.01, when Olig2 progenitors have normally switched to OPC produc- n ϭ 5; Fig. 2D–F). Moreover, as observed previously for Olig2, tion. Because OPCs, even reduced in number, were generated Sox10 signal intensity in Olig2 progenitors of sulf1Ϫ/Ϫ embryos from E11.5 in sulf1 mutant embryos (Fig. 2), these results sug- appeared weaker compared with wild-types at all stages examined gested that Olig2 progenitors might simultaneously generate (Fig. 2, compare D, E). High expression levels of these proteins in MNs and OPCs in sulf1-deficient embryos. According to the MN/ Olig2 progenitors being required to give them an OPC identity OPC switch model, this situation should never be observed in (Stolt et al., 2004; Liu et al., 2007), our results suggest that lowered wild-type conditions. To further investigate this point, we exam- Olig2/Sox10 expression might contribute to the phenotype ob- ined MN generation at the cellular level. For this, we analyzed served in sulf1 mutant embryos. Olig2 and Islet1 coexpression at various developmental stages in The marked reduction in the number of Olig2 progenitors that wild-type embryos and compared these patterns with the profile upregulate Sox10 in sulf1Ϫ/Ϫ embryos is consistent with Sulf1 acting obtained in E12.5 sulf1 mutant embryos. At E10.5 in wild-type by promoting OPC specification in the pMN domain. embryos, we detected Olig2/Islet1-coexpressing cells leaving the progenitor zone, corresponding to newly generated MNs that Sulf1 controls the MN-to-OPC fate switch of upregulated Islet1 before switching off Olig2 (Fig. 4A). At E11.5, Olig2 progenitors few Islet1-expressing cells were still detected in the medial region Because progenitors of the pMN domain generate MNs before of the spinal cord, close to the Olig2 progenitor domain, but they switching to the OPC fate, we next examined the consequences of no longer coexpressed Olig2, reflecting their more mature state Sulf1 inactivation on MN generation. At E12.5, immunostaining (Fig. 4B). Moreover, at this stage, we detected few Olig2-positive/ of Islet1 allowed detecting MNs in the ventral horns of the spinal Islet1-negative cells that had already left the progenitor zone, as cord, with no discernible difference between wild-type and well as Sox10-expressing cells in the progenitor zone (Fig. 4B,C), sulf1Ϫ/Ϫ embryos, indicating no gross defect in MN generation in indicating initiation of OPC generation. At E12.5, in agreement sulf1 mutants (Fig. 3A,B). However, we observed that, in sulf1Ϫ/Ϫ with the previously reported cessation of MN generation at this embryos, numerous Islet1-positive cells remained in a more me- stage (Stolt et al., 2003), none of the few Islet1-positive cells re- dial region, close to the Olig2 progenitor domain compared with maining in close vicinity to the pMN domain coexpressed Olig2 wild-type siblings, suggesting that MNs recently generated were in wild-type embryos (Fig. 4E). These data reinforce the model of overrepresented in sulf1 mutant embryos. To define whether MN the MN-to-OPC fate switch in mouse occurring in a narrow time production was indeed prolonged in sulf1Ϫ/Ϫ embryos, we first window during spinal cord development. In contrast, double used ebf2 and ngn2 stainings to identify young neurons in the Olig2/Islet1-positive young MNs were invariably detected in spinal cord, because their transitory expression in newly specified contact with the pMN domain until at least E12.5 in sulf1 mutant neurons is a reliable indicator of neuronal generation (Garel et embryos (Fig. 4F), confirming that MN production was pro- al., 1997; Dubois et al., 1998; Scardigli et al., 2001; Danesin et al., longed in this context, whereas few OPCs have already been spec- 2006). At E12.5, detection of ebf2 mRNA showed high level of ified (Fig. 4D). Therefore, in sulf1 mutant embryos, MNs and OPCs were simultaneously generated, a situation never encoun- 4 tered in wild-type embryos. This result is an additional argument against the possibility that the phenotype we observed in sulf1 (Figure legend continued.) Immunodetection of Olig2 on transverse sections of E12.5 (D–G), mutant embryos might simply result from a general delay in cell E14.5 (I–L), and P2 (N, O) spinal cords of wild-type (D, F, I, K, N) and sulf1Ϫ/Ϫ (E, G, J, L, O) fate change of Olig2 progenitors. embryos. D, E, Only few Olig2-positive cells emigrate in the mantle zone of E12.5 sulf1 mutant Ϫ/Ϫ embryos(E)comparedwithwild-typelittermates(D).F,G,HighermagnificationofE12.5spinal Together, our results show that sulf1 mutants display a cordsectionsinwhichtheprogenitorzoneisdelineatedbydashedlines.ArrowsindicateOlig2- partial and very specific failure of Olig2 progenitors to turn off positivecellshavingemigratedinthemantlezone.H,QuantificationofOlig2-expressingcellsin the MN program and switch on OPC production, resulting in a progenitorandmantlezonesofE12.5wild-type(ϩ/ϩ),sulf1heterozygous(ϩ/Ϫ),andsulf1 situation in which pMN progenitors concomitantly generate mutant(Ϫ/Ϫ)embryos.I,J,AtE14.5,thereductioninthenumberofmigratingOPCswasstill MNs and OPCs. This phenotype agrees with Sulf1 being involved apparent. K, L, Higher magnification of E14.5 spinal cord sections in which the progenitor zone in favoring the recruitment of Olig2 neural progenitors to gener- isdelineatedbydashedlines.NotethatthenumberofOlig2-positivecellsintheprogenitorzone ate OPCs instead of MNs. is reduced in sulf1 mutant embryos (L) compared with wild-type littermates (K). M, Quantifi- ϩ ϩ cation of Olig2-positive cells in progenitor and mantle zones of E14.5 wild-type ( / ) and Sulf1 activity is precisely required at the time of the MN-to- sulf1 mutant (Ϫ/Ϫ) embryos. N, O, At P2, Olig2-expressing cells populate the entire spinal OPC fate switch to trigger Olig2 progenitor cell fate change cordandarepresentinsimilardensitiesinwild-type(N)andsulf1mutant(O)embryos.P,Q,P2 spinal cord sections from wild-type (P) and sulf1 mutant (Q) embryos stained with the PLP/ The MN-to-OPC fate switch defect in sulf1-deficient mice raised DM20 probe. In H and M, cell numbers obtained for wild-type embryos were arbitrarily set to the question of the temporal window of Sulf1 activity in control- 100%.Allothervalueswereexpressedrelativetowild-typelevelsandarepresentedasmeanϮ ling the fate of Olig2 progenitors. Sulf1 expression pattern SEM. Statistical significance is indicated above the respective bars (*p Յ 0.02; ***p Յ 0.001). changes with time, and a novel source of the enzyme forms in Scale bars: A–E, I, J, N–Q, 100 ␮m; F, G, K, L,50␮m. ventral neural progenitors before the MN-to-OPC fate switch in Touahri et al. • Sulf1, An Activator of Shh-Dependent OPC Fate Choice J. Neurosci., December 12, 2012 • 32(50):18018–18034 • 18023

interference system described previously (Das et al., 2006). A sulf1RNAi vector was constructed and electroporated into ventral neural progenitors at the time of plating (E4.5). We first established its efficacy in downregulating Sulf1 expression using immunodetection and ISH after2din culture. Electroporation of the sulf1RNAi vector efficiently silenced Sulf1 expression at both protein and mRNA levels, whereas the use of a control vector ( gfpRNAi vector) had no effect (Fig. 5B,C and data not shown). We then examined the consequences of late inactivation of Sulf1 on OPC development. After2dinculture, electroporation of the sulf1RNAi vector led to a greater than twofold reduction in the number of Olig2-positive migrating OPCs in the mantle zone of the electroporated side (3.6 Ϯ 0.33, n ϭ 5) compared with the non-electroporated side of the explants (8.14 Ϯ 2.5, n ϭ 5; Fig. 5F,G), whereas electroporation of the control gf- pRNAi vector did not significantly affect this number (9.9 Ϯ 0.69, n ϭ 4) compared with the non-electroporated side of the explants (11,43 Ϯ 0.4, n ϭ 4; Fig. 5E,G)or non-electroporated explants (11.55 Ϯ 1.38 and 11.78 Ϯ 0.65 on either side of explants, n ϭ 5; Fig. 5A,G). To confirm these data, we used two distinct approaches reported previously to successfully impair Sulf1 function: overexpression of a dominant-negative form of Sulf1 (Sulf1 DN; Dhoot et al., 2001) or treatment with a Sulf1 blocking antibody (␣Sulf1; Gill et al., 2010). Similar results were obtained with both Figure 2. Sulf1 loss-of-function results in defective OPC specification from Olig2 progenitors. Double immunodetection of Sox10 and methods. Sulf1 DN overexpression caused a Olig2inE11.5(A,B)andE12.5(D,E)wild-type(A,D)andsulf1mutant(B,E)embryos.Imagesshowhighmagnificationsoftheprogenitor 50% reduction in the number of Olig2- zone, and each vertical set presents successively Sox10 staining, Olig2 staining, and the merged image (Sox10 in red and Olig2 in green). expressing cells in the mantle zone com- Note the reduction in the number of Olig2 and Sox10-coexpressing cells, as well as low Olig2 and Sox10 signal intensities in the few OPCs pared with the non-electroporated side of detectedintheprogenitorzoneofE12.5sulf1mutantembryoscomparedwithwild-typelittermates.C,QuantificationofSox10-expressing Ϯ Ϯ ϩ ϩ Ϫ Ϫ the explants (4.2 1.88 vs 9 1.50 cells, cellsintheprogenitorzoneofE11.5wild-type( / )andsulf1mutant( / )embryos.F,QuantificationofSox10-expressingcellsinthe ϭ ϩ ϩ Ϫ Ϫ n 4; Fig. 5J,K). Furthermore, treat- progenitor zone and in the mantle zone of E12.5 wild-type ( / ) andsulf1mutant ( / ) embryos. InCandF, cell numbers obtained ␣ for wild-type embryos were arbitrarily set to 100%. All other values were expressed relative to wild-type levels and are presented as ment with the Sulf1 blocking antibody mean ϮSEM. Statistical significance is indicated above the respective bars (*pՅ0.02; **pՅ0.01). Scale bars, 50 ␮m. resulted in a dose-dependent inhibition of OPC generation (Fig. 5L–N), with a 60 and 80% reduction in OPCs after in- both mouse and chicken (Fig. 1; Braquart-Varnier et al., 2004). cubation in 1:100 (4 Ϯ 0.37, n ϭ 10) and 1:50 (2.45 Ϯ 0.66, n ϭ This raises the question of the relevance of this temporal change 4) dilution of the ␣Sulf1 serum, respectively, compared with in Sulf1 expression pattern for proper OPC specification. To ad- control experiments performed with preimmune serum dress this question, we took advantage of the chicken model, (11.17 Ϯ 1.8, n ϭ 3). These data showed that Sulf1 activity is which allows interfering with gene function at late developmental required at the time of the MN-to-OPC fate switch for proper stages in explant cultures of embryonic spinal cord (Agius et al., OPC generation. 2004, 2010; Danesin et al., 2006). Chicken spinal cord explants Noticeably, in sulf1RNAi experiments mentioned above, were isolated at E4.5, i.e., just before OPC specification, which knockdown of sulf1 targeted to ventral neural progenitors but occurs between E5 and E5.5 (Soula et al., 2001). This stage also also to floor plate cells did not modify the number of OPCs in the corresponds to initiation of Sulf1 expression in ventral neural non-electroporated side of explants compared with control con- progenitors in chicken (Braquart-Varnier et al., 2004). With this ditions (Fig. 5F). This observation suggested that Sulf1 function system, the temporal fate of neural progenitors is maintained and for OPC generation could be fulfilled by ventral neural progeni- OPC migration in the mantle zone can be readily observed after tors but not by floor plate cells. We therefore examined spinal 2 d in culture, a stage equivalent to E6.5 (Agius et al., 2004; Dane- cord explants in which electroporation of the sulf1RNAi vector sin et al., 2006) (Fig. 5A). As a first step in the use of this ex vivo was mainly targeted to ventral medial cells and observed that the experimental model to study Sulf1 function, we adapted an RNA number of migrating OPCs on either side of the spinal cord ex- 18024 • J. Neurosci., December 12, 2012 • 32(50):18018–18034 Touahri et al. • Sulf1, An Activator of Shh-Dependent OPC Fate Choice

plant was not modified compared with non-electroporated control explants (Fig. 5H,I), confirming that expression of Sulf1 in floor plate cells is dispensable for OPC generation. To assess whether sulf1 knockdown in chicken also affects OPC specification, we performed double staining of Olig2 and Nkx2.2, their coexpression in neural progenitors being a well-established hallmark of OPC specification in chicken (Zhou et al., 2001). Impairment of Sulf1 function using the ␣Sulf1 blocking antibody led to a reduction of 60% in the number of Nkx2.2/Olig2-coexpressing cells in the progenitor zone (2.9 Ϯ 0.22 cells, n ϭ 4, p Ͻ 0.0001) compared with control explants incubated with the preimmune serum (7.53 Ϯ 0.44, n ϭ 3; Fig. 6A,B). These results indicated that Sulf1 is required for upregulation of Nkx2.2 expression in Olig2 progenitors to trigger their cell fate switch. In addition, we observed a clear defect in the dorsal extension of the Nkx2.2-positive/Olig2-negative p3 domain after Sulf1 inactivation (Fig. 6AЈ,BЈ), a fact very reminiscent of Shh signaling inhibition (see below). We then assessed MN production using the specific marker MNR2 in spinal cord explants either treated with the ␣Sulf1 blocking antibody or electroporated with the sulf1RNAi vector. We observed an increase in MN density close to the progenitor zone in explants incubated with the ␣Sulf1 antibody compared with control explants incubated with the preimmune serum (Fig. 6C,D). Similarly, after electroporation of the sulf1RNAi vector, MNR2-positive cells in immediate vicinity of the progenitor zone were overrepresented in the electroporated side of explants compared with the non- electroporated side (Fig. 6G). These results suggested that inactivation of Sulf1 at stage of the MN-to-OPC fate switch leads to pro-

sulf1 mutant (B) embryos. Note the detection of Islet1- expressing cells close to the Olig2 progenitor domain in E12.5 sulf1 mutant embryos (B, arrow), whereas only few of them were found in wild-type littermates (A, arrow). C–F, Double labeling of E12.5 spinal cord sections using ISH for ebf2 (C, D) or ngn2 (E, F) probes and immunodetection for Olig2 (green), performed on wild-type (C, E) and sulf1 mutant (D, F) embryos. In C, D, E, and F, a confocal image of Olig2 staining (green) was overlaid with the bright-field image of the same section to allow precise positioning of the pMN domain relative to cells expressing ebf2 (C, D) and ngn2 (E, F). Expression profiles of ebf2 and ngn2 appeared quite similar between wild-type and sulf1 mutant embryos except at the level of the pMN domain (delineated with brackets), in which a higher level of ebf2 expression (compare C؅ with D؅) and ngn2- -Figure 3. Generation of neurons is prolonged after the normal timing of the MN-to-OPC fate switch in sulf1-deficient mouse positive cells (compare E؅ with F؅) were detected in sulf1 mu embryos. A, B, Double immunodetection of Islet1 (red) and Olig2 (green) on transverse sections of E12.5 wild-type (A) and tant embryos. Scale bars, 50 ␮m. Touahri et al. • Sulf1, An Activator of Shh-Dependent OPC Fate Choice J. Neurosci., December 12, 2012 • 32(50):18018–18034 • 18025

conditions, rare BrdU- and MNR2-positive cells were detected in explants incubated with BrdU at 24 h (0.21 Ϯ 0.1, n ϭ 3), and they were no longer detected when pulses were performed at 32 h (n ϭ 3) and 48 h (n ϭ 3; Fig. 6E), in agreement with cessation of MN generation after E5/E5.5 in chicken. In contrast, double-labeled cells were invariably observed in explants treated with the ␣Sulf1 blocking antibody. Cell counting showed a number of 1.24 Ϯ 0.2, 1.19 Ϯ 0.1, and 0.47 Ϯ 0.1 double-labeled cells for pulses performed on at least three distinct explants at 24, 32, and 48 h (Fig. 6F and data not shown), respectively, confirming prolonged generation of MNs from actively proliferating cells during the last 24 h of culture, i.e., until E6.5. These results show that inactivation of Sulf1 just before the MN-to- OPC fate switch is sufficient to prevent ar- rest of MN generation by Olig2 progenitors. Together, these results confirm and ex- tend data obtained in mouse by showing that Sulf1 activity is required at the time of the MN-to-OPC fate switch to trigger Olig2 progenitor cell fate change.

Sulf1 is a positive regulator of Shh signaling in Olig2 progenitors as they switch to the OPC fate Our functional analyses implying Sulf1 in the control of the MN-to-OPC fate switch open the question of the mechanism involved. Hedgehog signaling, given its central role in OPC specification and its functional relationship with Sulf1 both in chicken and Drosophila (Danesin et al., 2006; Wojcinski et al., 2011), was proposed as a prime candidate to be regulated by Sulf1. Our previous data (Danesin et al., 2006), showing that overexpression of Sulf1 in the developing chicken spinal cord is sufficient to upregulate ptc1, the Shh receptor and also an Shh-responsive gene (Goodrich et al., 1996), indicated that Sulf1 behaves as a positive regulator of Shh in this tissue. To further evaluate the role of Sulf1 in controlling the Shh Figure4. MNsandOPCsaresimultaneouslygeneratedinE12.5sulf1mutantembryos.A,B,TimecourseofIslet1(red)andOlig2(green) pathway at the MN-to-OPC switch, we -expression in E10.5 (A) and E11.5 (B) wild-type embryos. A؅, B؅, Higher magnifications of the areas framed in A and B, showing newly first analyzed the expression of Shh .generatedMNsmarkedbythecoexpressionofbothtranscriptionfactorsatE10.5(arrowsinA؅)butnotatE11.5(B؅).C–F,Comparisonof responsive genes after Sulf1 inactivation Islet1,Sox10,andOlig2expressioninwild-type(C,E)andsulf1mutant(D,F)embryos.C,D,DoubleimmunodetectionofIslet1(green)and As reported previously (Danesin et al., Sox10 (red) in E11.5 wild-type (C) and sulf1 mutant (D) embryos showing higher density of Islet1-expressing cells close to the progenitor 2006), between E4 and E6 in chicken, ptc1 zoneinsulf1mutantembryos.NotethepresenceoffewSox10-positiveOPCsalreadydetectedatthisstageinsulf1mutantspinalcord.E,F, expression became elevated in two bilat- Double staining of Islet1 (red) and Olig2 (green) performed on wild-type (E) and sulf1 mutant (F) E12.5 spinal cords. E؅, F؅, Higher eral domains of the ventral progenitor magnificationsoftheareasframedinEandFshowingthatcellscoexpressingOlig2andIslet1(arrowsinF؅)weredetectedinsulf1mutant zone (Fig. 7A), indicating increased level .embryos but not in wild-type littermates (E؅). Scale bars: A,B,C–F,50␮m; A؅,B؅,E؅,F؅,20␮m. of Shh activity in these neural progenitors Addition of Sulf1 blocking antibody at the longed production of MNs from Olig2 progenitors, as observed in time of plating led to inhibition of ptc1 expression at this level sulf1-deficient mouse embryos. To confirm this, we performed (Fig. 7B), indicating that Sulf1 inactivation prevented the rise in birthdating experiments using BrdU. Two hour BrdU pulses were Shh signaling normally observed during this time window. In the performed at 24, 32, and 48 h after plating, equivalent stages to E5.5, mouse, whether or not a similar temporal elevation of Shh activity is E6, and E6.5, respectively (Fig. 6E,F and data not shown). In control also responsible for the MN-to-OPC fate switch is unknown. To 18026 • J. Neurosci., December 12, 2012 • 32(50):18018–18034 Touahri et al. • Sulf1, An Activator of Shh-Dependent OPC Fate Choice

Figure 5. Impairment of Sulf1 function immediately before the MN-to-OPC fate switch is sufficient to inhibit OPC generation in chicken spinal cord. A, Immunodetection of Olig2 on a transverse section of chickenspinalcordsisolatedatE4.5andcultivatedasopened-bookexplantsshowingOPCsgeneratedafter2dincultureincontrolconditions.Thedashedlinesdelineatetheprogenitorzone(PZ)andthemantle zone(MZ).NotedetectionofOPCsmigratinginthemantlezoneoneachsideofspinalcordcontrolexplants.B,C,ImmunodetectionofSulf1afterelectroporationofgfpRNAicontrolvector(B)andsulf1RNAivector (C).Becausetheuseofethanol/acetic-acidfixative,requiredforpreservationofSulf1immunoreactivity,alsoleadstoextinctionoftheRFPfluorescentsignal,electroporationismonitoredbeforefixationonwhole spinalcordexplantsasexemplifiedinD(dashedlineindicatesthemidline,RFPsignalisinblue).InBandC,theelectroporatedsideofexplantsisontheright.NotedetectionofSulf1infloorplatecells(FP),aswell asinbilateralventralneuralprogenitorsinexplantselectroporatedwithgfpRNAicontrolvector(B),whereaselectroporationwithsulf1RNAivectorinhibitsSulf1immunoreactivityontheelectroporatedsideof theexplants(asteriskinC).E,E؅,ElectroporationofthegfpRNAicontrolvector(blueinE؅)doesnotaffectgenerationofOlig2-positiveOPCs(redinE؅)ontheelectroporatedsideoftheexplantscomparedwith the non-electroporated contralateral side or to control explants (compare E with A). F, F؅, Electroporation of the sulf1RNAi vector (blue in F؅) in a broad domain covering the ventral progenitor zone inhibits generationofOlig2-positiveOPCsinthemantlezone(redinF؅)intheelectroporatedsideoftheexplantscomparedwiththenon-electroporatedcontralateralsideorwithcontrolexplants(compareleftsidein FandA).H,H؅,Electroporationofthesulf1RNAivector(blueinH؅)inventralmedialcellsbutnotinneuralprogenitorsadjacenttotheOlig2progenitordomain(asterisksinH؅)hasnoeffectontheproduction ofOlig2-positiveOPCs(redinH؅)comparedwithnon-electroporatedcontrolexplants(compareHwithA).J,J؅,ElectroporationoftheSulf1 DN-expressingvector(blueinJ؅)inhibitsgenerationofOlig2-positive OPCs(redinJ؅)ontheelectroporatedsideoftheexplantscomparedwiththenon-electroporatedsideandwithcontrolexplants(compareJandA).G,I,K,QuantificationofOlig2-positivecellsmigratinginthe mantle zone on each side of the explants in control condition (control), after electroporation of gfpRNAi control vector (G), sulf1RNAi vector (G, I), and Sulf1 DN-expressing vector (K). Results are presented as meanϮSEMnumberofcells(***pՅ0.001).ne,Non-electroporatedsideoftheexplants;e,electroporatedsideoftheexplants.L,M,Olig2immunolabelingaftertreatmentwiththecontrolpreimmuneserum (controlPI,L)orSulf1-blockingantibody(␣Sulf1)dilutedat1:50(M).N,QuantificationofOlig2-positivecellsmigratinginthemantlezoneofspinalcordexplantscultivatedincontrolcondition(control),inthe presence of the preimmune serum (control PI), or after incubation with ␣Sulf1 diluted at 1:100 and 1:50. Results are presented as mean ϮSEM number of cells per optical hemisection of spinal cord explants (**pՅ0.01). FP, Floor plate. Scale bars, 50 ␮m. Touahri et al. • Sulf1, An Activator of Shh-Dependent OPC Fate Choice J. Neurosci., December 12, 2012 • 32(50):18018–18034 • 18027

littermates. Detection of ptc1 mRNA in wild-type embryos showed differential levels of expression between the dorsal and ventral parts of the progenitor zone, with the signal being stronger in ventral progenitors (Fig. 7C,E), indicating a higher Shh activity in this region. In contrast, we observed a quite homogeneous signal all along the dorsoventral axis in sulf1 mutant spinal cords (Fig. 7F). A strong expression of gli1 was also observed in the ventral region of wild-type embryos (Fig. 7D,G). In contrast, in sulf1 mutant littermates, only faint expression of gli1 was detected at this level, whereas the more dorsal expression domains seemed unaffected (Fig. 7H). These results, in good agreement with a high level of Shh signaling in ventral neural progenitors at the onset of OPC generation in mouse embryos, support a positive regulatory role of Sulf1 in activating Shh signaling at this stage. Importantly, we found that shh expression was not modified in sulf1 mutant embryos compared with wild-type siblings (Fig. 7I,J), indicating that the influence of Sulf1 on Shh signaling did not depend on transcriptional downregulation of the morphogen factor. To confirm the functional relationship between Sulf1 and Shh in mouse, we turned to genetic interaction experiments. We postulated that, if the reduction in OPC generation in sulf1Ϫ/Ϫ embryos indeed re- flected a reduced level of Shh signaling in Olig2 progenitors, then lowering the dosage of shh would be expected to enhance this phenotype. To test this prediction, sulf1;shh compound mutants were analyzed. First, we compared wild-type and shhϩ/Ϫ littermate embryos and found that shhϩ/Ϫ E12.5 embryos were phenotypically normal and Figure 6. Impairment of Sulf1 function during the MN-to-OPC fate switch inhibits OPC specification and allows prolonged MN production in chicken spinal cord. A–B؅, Double detection of Olig2 (red) and Nkx2.2 (green) after treatment with the preimmune displayed a number of Olig2/Sox10- serum (control PI, A) or Sulf1 blocking antibody diluted at 1:50 (␣Sulf1, B). Treatment with ␣Sulf1 reduced the number of coexpressing OPCs in the progenitor zone Olig2/Nkx2.2 double-labeled OPCs in both the mantle zone and the progenitor zone (delineated by dashed line) compared with similar to that of wild-type littermates ϩ/Ϫ control explants incubated with the preimmune serum. Note reduction of the dorsoventral extension of the Nkx2.2-expressing (22.27 Ϯ 0.65 in shh embryos, n ϭ 3, vs ;domain (brackets in A؅, B؅) in explants incubated with ␣Sulf1 (B؅) compared with control PI explants (A؅). C–F, Detection of 22.26 Ϯ 1.77 in wild-type siblings, n ϭ 4 Olig2 (red), MNR2 (green), and BrdU (blue in E, F) in control PI conditions (C, E) or in the presence of ␣Sulf1 (D, F). E and Fig. 8A–C). Similarly, loss of one shh allele F show higher magnification of the progenitor zone (PZ). Note that MNR2-labeled cells in contact with the Olig2-positive had no effect on OPCs migrating in the domainareoverrepresentedintheexplantstreatedwith␣Sulf1(arrowsinD,greeninF)comparedwithcontrolPIexplants mantle zone (22.2 Ϯ 1.2, n ϭ 3inshhϩ/Ϫ (C, E). In E and F, a BrdU pulse has been performed at 32 h after plating. Note the presence of MNR2-positive cells stained Ϯ ␣ embryos vs 23 1.12 in wild-type siblings, with the BrdU antibody in explants treated with Sulf1 (arrows in F) but not in control PI explants (E). G, Double detection n ϭ 4; Fig. 8C and data not shown). We of Olig2 (red) and MNR2 (green) on explants electroporated with the sulf1RNAi vector (blue) showing accumulation of next compared OPC specification in MNR2-positive cells on the electroporated side (arrow) compared with the non-electroporated side. FP, Floor plate. Scale Ϫ/Ϫ ϩ/Ϫ Ϫ/Ϫ ϩ/ϩ bars: A–D, G,50␮m; E, F,20␮m. sulf1 ;shh and sulf1 ;shh E13.5 littermate embryos. The number of Olig2 and Sox10-coexpressing OPCs evaluate Shh signaling in mouse embryonic spinal cord, we analyzed in the progenitor zone was significantly reduced in sulf1Ϫ/Ϫ; expression of ptc1 and gli1, the latter being reported to be activated in shhϩ/Ϫ embryos (7.59 Ϯ 0.04, n ϭ 4) compared with sulf1Ϫ/Ϫ; cells that are exposed to high levels of Shh signaling (Bai et al., 2002; shhϩ/ϩ littermates (13.3 Ϯ 0.52 cells, n ϭ 4; Fig. 8D–F), showing Bai et al., 2004). Double detection of each of these genes together a genetic interaction between sulf1 and shh. with Olig2 at E12.5 showed both ptc1 and gli1 mRNA expression in Together, our data evidence a role for Sulf1 in controlling Shh the pMN domain (Fig. 7C,D), indicating that Olig2 progenitors are signaling at the time of OPC specification in both chicken and indeed subjected to Shh activity at this stage. We then compared the mouse, with the enzyme acting as a positive regulator of this expression patterns of ptc1 and gli1 in E12.5 wild-type and sulf1Ϫ/Ϫ signal. 18028 • J. Neurosci., December 12, 2012 • 32(50):18018–18034 Touahri et al. • Sulf1, An Activator of Shh-Dependent OPC Fate Choice

Sulf1 contributes to Nkx2.2 upregulation in Olig2 progenitors as they switch to OPC fate in mouse In chicken, a temporal change in dorsoventral patterning of the spinal cord has been reported to occur just before the MN-to-OPC fate switch. From E5 on- ward, the Nkx2.2 progenitor domain ex- tends dorsally, within the pMN domain, resulting in formation of the p* domain characterized by the coexpression of Olig2 and Nkx2.2 (Zhou et al., 2001; Fu et al., 2002; Agius et al., 2004). We reported previously that this upregulation of Nkx2.2 in Olig2 progenitors depends on Shh (Agius et al., 2004). Our present results showing impairment of Nkx2.2 domain extension after inactivation of Sulf1 in chicken spinal cord explants (Fig. 6AЈ,BЈ) bring additional support to a role for Sulf1 in activating Shh signaling at the time of the MN-to-OPC fate switch. In mouse embryos, coexpression of Olig2 and Nkx2.2 in ventral neural progenitors at the stage of OPC specification has been reported previously (Sun et al., 2001), and OPCs are also well known to upregulate Nkx2.2 expression while they migrate in the mantle zone (Qi et al., 2001; Fu et al., 2002). We performed double immunodetection experiments at various stages of development and observed that, in wild-type mouse embryos, at least a subpopulation of Olig2 progenitors in the pMN domain gained Nkx2.2 expression between E10.5 and E12.5 (Fig. 9A,B). However, the Nkx2.2 signal intensity in Olig2 progenitors was quite weak, reflecting a low level of expression. At E12.5, we found that 55% of Olig2 progenitors coexpressed Nkx2.2 (23.95 Ϯ 1.36, n ϭ 4). In sulf1Ϫ/Ϫ embryos, the number of Olig2/Nkx2.2- coexpressing cells in the progenitor

Figure 7. Sulf1 positively regulates expression of Shh target genes in the ventral embryonic spinal cord at the onset of OPC generation. A, B, Transverse sections of chicken spinal cord explants isolated at E4.5, maintained in culture for2din the presence of preimmune serum (control PI, A) or Sulf1 blocking antibody (␣Sulf1, B), and processed for detection of ptc1 mRNA. Note decreased expression of ptc1 during treatment with ␣Sulf1 (B) compared with control PI (A). C–J, Transverse spinal cord sections of E12.5 mouse embryos. C, D, ISH performed on wild-type embryos using ptc1 (C) and gli1 (D) probes, followed by immunodetection of Olig2. Confocal imagesofOlig2staining(green)wereoverlaidwiththebright- field image of the same section to position the pMN domain relativetocellsexpressingptc1andgli1.E–J,Detectionofptc1 (E,F),gli1(G,H)andshh(I,J)mRNAsonwild-type(E,G,I)and sulf1 mutant (F, H, J) embryos. Note decreased ptc1 and gli1 expressionsintheventralprogenitorzoneofsulf1mutantem- bryos (F, H) compared with wild-type embryos (E, G). Scale bars: A–D,50␮m; E–J, 100 ␮m. Touahri et al. • Sulf1, An Activator of Shh-Dependent OPC Fate Choice J. Neurosci., December 12, 2012 • 32(50):18018–18034 • 18029

expressing Sulf1 relative to Olig2 progenitors. For this, we performed double detection of Sulf1 and Olig2 at E12.5 in mouse and E6 in chicken. Our results showed that, in both species, Sulf1- expressing cells were located in the ventralmost region of the progenitor zone but were excluded from the Olig2-expressing domain (Fig. 10A,B). Therefore, Olig2 progenitors do not express Sulf1, opening the possibility that the enzyme induces their switch to the OPC fate by non-cell autonomously enhancing Shh activity. To investigate this possibility, we first examined expression of Shh-responsive genes relative to sulf1 domain of expression by double labeling for sulf1 and ptc1 or gli1 in E6 chicken spinal cord. Our results showed that the highest levels of ptc1 and gli1 expression were detected in progenitors located dorsally to sulf1-expressing cells (Fig. 10C,D), in agreement with the possibility that Sulf1 would stimulate the response to Shh in dorsally adjacent cells. We next performed Sulf1 overexpression experiments using electroporation of a Sulf1-expressing vector in E4.5 chicken spinal cord. Misexpression of Sulf1 in- duced upregulation of ptc1 after2dincul- ture, whereas electroporation of an empty vector did not modify its expression (Fig. 10E,F). Importantly, cells displaying high levels of ptc1 expression were located dorsally to and excluded from the ventralmost domain containing a high density of cells electroporated with the sulf1- expressing vector (Fig. 10E). To deter- Figure8. sulf1;shhcompoundmutantsdisplayseverelyreducedOPCspecification.A,B,DoubledetectionsofOlig2(greeninA؅, mine whether ventrally localized Sulf1 B؅) and Sox10 (red in A؅, B؅) in the progenitor zone of E12.5 wild-type (A, A؅) and shhϩ/Ϫ (B, B؅) embryos. C, Quantification of ϩ/Ϫ overexpression may be sufficient to en- Olig2/Sox10-coexpressing cells in wild-type and shh embryos, in both the progenitor zone and the mantle zone. D, E, Double hance OPC generation, we analyzed detections of Olig2 (green in D؅, E؅) and Sox10 (red in D؅, E؅) in the progenitor zone of E13.5 sulf1Ϫ/Ϫ;shhϩ/ϩ (D, D؅) and Ϫ Ϫ ϩ Ϫ Olig2 expression after electroporation sulf1 / ;shh / (E, E؅) littermate embryos. F, Quantification of Olig2/Sox10-coexpressing cells in the progenitor zone in the different genetic contexts indicated under bars. Cell numbers obtained for wild-type (C) and sulf1Ϫ/Ϫ;shhϩ/ϩ (F) embryos were of the sulf1-expressing vector. The num- arbitrarily set to 100%. All other values were expressed relative to wild-type (C)orsulf1Ϫ/Ϫ;shhϩ/ϩ (F) levels and are presented ber of OPCs migrating in the marginal as mean Ϯ SEM (*p Յ 0.02). WT, Wild type. Scale bars, 50 ␮m. zone was markedly elevated in the electroporated side compared with the contralateral side (n ϭ 5) or with control zone was reduced with only 32% of Olig2 progenitors (14 Ϯ ϭ ϭ ϭ explants electroporated with an empty vector (n 3; Fig. 1.22, n 4, p 0.02) having upregulated Nkx2.2 at this stage 10G,H). These results indicate that neural progenitors over- (Fig. 9C). expressing Sulf1 gain the ability to elevate Shh activity in dor- Together, these results indicate that sulf1 impairment caused sally located cells and stimulate OPC generation from Olig2 a 40% reduction in the number of Olig2/Nkx2.2-coexpressing progenitors. cells. Therefore, as observed in chicken, dorsal extension of Together, these data suggest a role for Sulf1 as an enhancer of Nkx2.2 expression in mouse embryos also depends on Sulf1 ac- Shh activity in immediately neighboring cells. tivity. Moreover, because Nkx2.2 is recognized as a direct target of Shh (Lei et al., 2006; Vokes et al., 2007; Lek et al., 2010), these Discussion results bring additional support to Sulf1 acting by positively reg- Olig2 progenitors of the embryonic spinal cord change their fate ulating Shh signaling activity. during a short developmental window from the production of MNs to that of OPCs. Here we report a novel function for Sulf1 as Sulf1 is not expressed in Olig2 progenitors and acts non-cell an enzyme involved in the control of this neuroglial switch in autonomously to positively regulate Shh responsive gene both mouse and chicken. expression We show that lack of Sulf1 activity leads to defective OPC To get additional insight on Sulf1 function in the ventral spinal generation. This does not result from an initial depletion in the cord, it was important to precisely position neural progenitors Olig2 progenitor pool, which is not affected in sulf1 mutant em- 18030 • J. Neurosci., December 12, 2012 • 32(50):18018–18034 Touahri et al. • Sulf1, An Activator of Shh-Dependent OPC Fate Choice

Figure 9. Nkx2.2 upregulation in Olig2 progenitors occurs at the stage of MN-to-OPC fate switch in mouse embryos and depends on Sulf1 function. A–C, Double detection of Nkx2.2 (A–C, red in A؆–C؆)andOlig2(A؅–C؅,greeninA؆–C؆)ontransversesectionsofE10.5(A)andE12.5(B)wild-typeandofE12.5sulf1mutant(C)embryos.NotedorsalextensionoftheNkx2.2-expressingdomain between E10.5 (A) and E12.5 (B) in wild-type embryos, which is reduced in sulf1 mutant embryos (compare B with C). Although only rare cells coexpress Olig2 and Nkx2.2 at E10.5 (A؆), a broad overlappingbetweenthesetwotranscriptionfactorswasvisualizedinE12.5wild-typeembryos(bracketinB؆).InE12.5sulf1mutantspinalcords,onlyfewcellscoexpressOlig2andNkx2.2(bracket .in C؆) compared with wild-type siblings (compare C؆ with B؆). Scale bars, 50 ␮m bryonic spinal cord at the onset of OPC production. Instead, this effect on OPC generation, pointing to the major contribution of defect relies on the reduction in the number of Olig2 progenitors sulf1 upregulated in Nkx2.2-expressing cells of the p3 domain that should have turned off the MN program to switch to the just before the onset of OPC generation (Braquart-Varnier et al., OPC fate. As a consequence, a remarkable phenotype in sulf1- 2004). We previously reported that, in chicken, OPC specifica- deficient spinal cords is the concomitant generation of MNs and tion depends on a rise in Shh activity occurring precisely at this OPCs from Olig2 progenitors, a situation never encountered in stage (Danesin et al., 2006). Our present results showing upregu- normal conditions, in which MN and OPC productions are lation of Nkx2.2 in Olig2 progenitors at the stage of the MN-to- timely separated events in both mouse and chicken (Kessaris et OPC fate switch in the mouse strongly argue in favor of a timely al., 2001; Soula et al., 2001; our present results). Therefore, Sulf1 regulated elevation of Shh activity also conserved in mice. Indeed, activity appears to be required for both cessation of MN produc- Nkx2.2 expression depends on a high level of Shh activity, and tion and optimal recruitment of Olig2 progenitors to generate this transcription factor has also been reported to intrinsically OPCs at the right time. strengthen Shh response in neural progenitors (Novitch et al., Experiments performed in chicken show that impairment of 2001; Lei et al., 2006; Dessaud et al., 2007; Lek et al., 2010). Our Sulf1 activity at the stage of the MN-to-OPC fate switch is suffi- data showing genetic interaction between sulf1 and shh in the cient to reproduce the phenotype observed in sulf1 mutant mouse, as well as downregulation of Shh-responsive genes in mouse embryos, supporting the view that Sulf1 activity is re- Olig2 progenitors after inactivation of Sulf1, establish a critical quired precisely when Olig2 progenitors change their fate. In role for the enzyme in controlling Shh activity in the developing support of this, knockdown of sulf1 in floor plate cells has no ventral spinal cord. Both upregulation and maintenance of olig2 Touahri et al. • Sulf1, An Activator of Shh-Dependent OPC Fate Choice J. Neurosci., December 12, 2012 • 32(50):18018–18034 • 18031

Figure10. Sulf1isnotexpressedinOlig2progenitorsandactsnon-cellautonomouslytopositivelyactivateShhsignalingactivity.A,B,TransversesectionsofE12.5mouse(A)andE6chicken(B) embryonic spinal cords, double immunolabeled for Olig2 (red) and Sulf1 (green). Sulf1 is detectable in floor plate (FP) cells as well as in the ventralmost neural progenitors of the p3 domain but not in dorsally located Olig2-expressing cells. C, D, Double detections of sulf1 (red) and ptc1 (purple in C)orgli1 (purple in D) mRNAs showing high level of expression for both Shh responsive genes in neural progenitors (brackets) located dorsally to sulf1-expressing cells. E–H, Transverse sections of chicken spinal cord explants isolated at E4.5 and maintained in culture for 2 d. E, F, Detection of ptc1 mRNA (purple) after electroporation of sulf1-expressing vector (green, E) or control empty vector (green, F). Note upregulation of ptc1 in neural progenitors located dorsally relatively to the domaincontaininghighdensityofsulf1-overexpressingcells(E).G,H,ImmunodetectionofOlig2(red)afterelectroporationwiththesulf1-expressingvector(blue,G)orcontrolemptyvector(blue, H)showingamarkedincreaseinthepopulationofOlig2-positivecellsmigratinginthemantlezoneofexplantselectroporatedwiththesulf1-expressingvector(G)comparedwiththecontralateral side of explants or with explants electroporated with the control vector (H). Scale bars: A, B, E–H,50␮m; C, D, 100 ␮m. expression in the ventral neural tube is well known to depend on phosphorylation status reported recently to guide its activity in Shh (Lu et al., 2000; Zhou et al., 2000; Agius et al., 2004). Because promoting the MN or OPC fate choice (Li et al., 2011). Our data the pool of Olig2 progenitors is not affected in E12.5 sulf1 mutant also open the possibility that Nkx2.2 might be part of the regula- embryos, the enzyme appears to act as a positive modulator but tory machinery that triggers the OPC fate choice in mouse, as not as an obligatory component of the Shh signaling pathway. reported in chicken (Zhou et al., 2001). Finally, the phenotype of Therefore, Sulf1 should be viewed as a temporal regulator of Shh sulf1 mutant embryos is very reminiscent of that observed after signaling in the developing spinal cord involved in elevating Shh ablation of Sox9 in neural progenitors of aged-matched embryos activity to trigger Olig2 progenitor cell fate change. (Stolt et al., 2003). Sox9 upregulates expression of the gliogenic One attractive hypothesis, supported by the lowered intensity factor NFIA (nuclear factor 1A), which further acts in coordina- of the Olig2 signal observed in the progenitor zone of sulf1 mu- tion with Olig2 to favor the OPC fate choice of Olig2 progenitors tant mouse embryos, is that Sulf1-dependent elevation of Shh (Deneen et al., 2006; Hochstim et al., 2008; Kang et al., 2012). might contribute to upregulate Olig2 expression in at least sub- Because sox9 expression in ventral neural progenitors is also reg- sets of pMN progenitors. Indeed, variations of Olig2 expression ulated by Shh (Scott et al., 2010), this transcription factor repre- levels in pMN cells have been reported previously to regulate MN sents another possible target of Sulf1-dependent timely regulated or OPC generation: downregulation of Olig2 is required to re- fate of Olig2 progenitors. lease the repressive block of this factor on neurogenesis, whereas Our data raise the question of the molecular mechanisms un- high Olig2 expression triggers progenitors to follow an OPC fate derlying the Sulf1-dependent temporal regulation of Shh activity. (Novitch et al., 2001; Qi et al., 2003; Liu et al., 2007). Another We know from our previous work that Sulf1 is involved in shap- possible level of regulation might be linked to the control of Olig2 ing the Shh/Hh gradient at the apical surface of chicken neural 18032 • J. Neurosci., December 12, 2012 • 32(50):18018–18034 Touahri et al. • Sulf1, An Activator of Shh-Dependent OPC Fate Choice progenitors and Drosophila wing disc cells (Danesin et al., 2006; Ext1 is upregulated in the ventral developing spinal cord precisely Wojcinski et al., 2011). Sulf1 is well known to remove 6O-sulfate at the stage of the MN-to-OPC fate switch in chicken (Danesin et groups on HS chains of HSPGs, the only known sulfate moiety al., 2006). post-synthetically edited (Lamanna et al., 2007). Accordingly, Overall, our work, by characterizing Sulf1 as a positive regu- loss of Sulf1 in mice selectively increases the abundance of trisul- lator of the MN-to-OPC fate switch, sheds new light on the fated disaccharides and proportionately decreases disulfated ones mechanisms underlying the timely regulated activation of Shh (Ai et al., 2006; Lamanna et al., 2006; Nagamine et al., 2012). signaling responsible for OPC specification at the expense of MN HSPGs are currently viewed as major regulators for morphogen production in the embryonic ventral spinal cord. gradient formation during development (Bishop et al., 2007), and Shh is well known to interact with HS chains, mainly by its References Cardin–Weintraub motif (Carrasco et al., 2005; Zhang et al., Agius E, Soukkarieh C, Danesin C, Kan P, Takebayashi H, Soula C, Cochard 2007; Chang et al., 2011; Farshi et al., 2011). Importantly, Shh is P (2004) Converse control of oligodendrocyte and astrocyte lineage de- known to preferentially bind to HS chains having a high level of velopment by Sonic hedgehog in the chick spinal cord. Dev Biol 270:308– sulfation (Carrasco et al., 2005; Zhang et al., 2007; Dierker et al., 321. CrossRef Medline 2009). Therefore, as reported for other signaling molecules (Ai et Agius E, Decker Y, Soukkarieh C, Soula C, Cochard P (2010) Role of BMPs in controlling the spatial and temporal origin of GFAP astrocytes in the al., 2003; Viviano et al., 2004; Freeman et al., 2008), Sulf1 activity embryonic spinal cord. Dev Biol 344:611–620. CrossRef Medline is likely to lower Shh/HSPGs interaction. Because Sulf1 is not Ai X, Do AT, Lozynska O, Kusche-Gullberg M, Lindahl U, Emerson CP Jr expressed in Olig2 progenitors themselves when these cells re- (2003) QSulf1 remodels the 6-O sulfation states of cell surface heparan spond to high Shh activity, i.e., E12.5 in mouse and E6 in chicken, sulfate proteoglycans to promote Wnt signaling. J Cell Biol 162:341–351. it is unlikely that Sulf1 acts locally. However, keeping in mind CrossRef Medline that the enzyme is secreted, the possibility that Sulf1 reaches the Ai X, Do AT, Kusche-Gullberg M, Lindahl U, Lu K, Emerson CP Jr (2006) surface of Olig2 progenitors cannot be totally excluded. A more Substrate specificity and domain functions of extracellular heparan sulfate 6-O-endosulfatases, QSulf1 and QSulf2. J Biol Chem 281:4969–4976. convincing explanation that fits well with our results is that Sulf1, CrossRef Medline by eliminating 6O-sulfate groups on HS chains, locally lowers Ai X, Kitazawa T, Do AT, Kusche-Gullberg M, Labosky PA, Emerson CP Jr Shh/HSPG interaction and thereby promotes Shh release from (2007) SULF1 and SULF2 regulate heparan sulfate-mediated GDNF sig- the surface of p3 progenitors. Subsequently, because of higher naling for esophageal innervation. Development 134:3327–3338. availability of the diffusible form of Shh free to travel dorsally, CrossRef Medline higher doses of the morphogen are provided to neighboring Allen BL, Song JY, Izzi L, Althaus IW, Kang JS, Charron F, Krauss RS, McMa- hon AP (2011) Overlapping roles and collective requirement for the co- Olig2 progenitors. Accordingly, in chicken spinal cord explants, receptors GAS1, CDO, and BOC in SHH pathway function. Dev Cell ventral cells overexpressing Sulf1 induce ectopic ptc1 upregula- 20:775–787. CrossRef Medline tion in neural progenitors located dorsally but not in Sulf1- Bai CB, Auerbach W, Lee JS, Stephen D, Joyner AL (2002) Gli2, but not Gli1, overexpressing cells themselves. Moreover, ptc1 and gli1 mRNAs is required for initial Shh signaling and ectopic activation of the Shh are undetectable in the ventralmost neural progenitors of E6 pathway. Development 129:4753–4761. Medline chicken spinal cord, indicating low or no Shh activity in Sulf1- Bai CB, Stephen D, Joyner AL (2004) All mouse ventral spinal cord patterning by hedgehog is Gli dependent and involves an activator function of expressing cell themselves, again in agreement with Sulf1 acting Gli3. Dev Cell 6:103–115. CrossRef Medline by favoring Shh release from these cells. The fact that p3 progen- Bilican B, Fiore-Heriche C, Compston A, Allen ND, Chandran S (2008) itors maintain Nkx2.2 as they gain Sulf1 expression (our present Induction of Olig2 precursors by FGF involves BMP signalling blockade results; Braquart-varnier et al., 2004) does not disagree with this at the Smad level. PLoS One 3:e2863. CrossRef Medline proposal because Shh, known to be required for switching on Bishop JR, Schuksz M, Esko JD (2007) Heparan sulphate proteoglycans nkx2.2 in ventral neural progenitors, is dispensable for its main- fine-tune mammalian physiology. Nature 446:1030–1037. CrossRef tenance later on (Agius et al., 2004; Allen et al., 2011). Medline Braquart-Varnier C, Danesin C, Clouscard-Martinato C, Agius E, Escalas N, Our present work, by characterizing Sulf1 as a positive regu- Benazeraf B, Ai X, Emerson C, Cochard P, Soula C (2004) A subtractive lator of the Shh-dependent Olig2 progenitor cell fate change, approach to characterize genes with regionalized expression in the glio- highlights a role for HSPGs in the temporal control of the MN- genic ventral neuroepithelium: identification of chick sulfatase 1 as a new to-OPC fate switch. These components of the extracellular ma- oligodendrocyte lineage gene. Mol Cell Neurosci 25:612–628. CrossRef trix have been reported previously to control the proliferative Medline activity of Shh in the mouse developing cerebellum (Rubin et al., Briscoe J, Novitch BG (2008) Regulatory pathways linking progenitor patterning, cell fates and neurogenesis in the ventral neural tube. Philos 2002; Chan et al., 2009). However, Shh/HSPG interactions have Trans R Soc Lond B Biol Sci 363:57–70. CrossRef Medline been found dispensable for achieving Shh-dependent patterning Cai J, Qi Y, Hu X, Tan M, Liu Z, Zhang J, Li Q, Sander M, Qiu M (2005) of the ventral neural tube (Chan et al., 2009). In agreement with Generation of oligodendrocyte precursor cells from mouse dorsal spi- the aforementioned conclusion, our data indicate that Sulf1 is nal cord independent of Nkx6 regulation and Shh signaling. Neuron also dispensable for formation of the ventral progenitor domains 45:41–53. CrossRef Medline that express Nkx2.2 and Olig2, indicating that at least the modu- Carrasco H, Olivares GH, Faunes F, Oliva C, Larraín J (2005) Heparan sulfate proteoglycans exert positive and negative effects in Shh activity. J Cell lation of 6O-sulfation on HSPGs is not critically required for Shh Biochem 96:831–838. CrossRef Medline patterning activity in the ventral neural tube. However, the in- Chan JA, Balasubramanian S, Witt RM, Nazemi KJ, Choi Y, Pazyra-Murphy volvement of Sulf1 in controlling Shh-dependent cell fate change, MF, Walsh CO, Thompson M, Segal RA (2009) Proteoglycan interac- long after establishment of the ventral neural tube patterning, tions with Sonic Hedgehog specify mitogenic responses. Nat Neurosci indicates that Shh signaling becomes dependent on HSPG regu- 12:409–417. CrossRef Medline lation as development proceeds. Because expression of HSPGs is Chandran S, Kato H, Gerreli D, Compston A, Svendsen CN, Allen ND tightly controlled during development (Yan and Lin, 2009), it is (2003) FGF-dependent generation of oligodendrocytes by a hedgehog- independent pathway. Development 130:6599–6609. CrossRef Medline tempting to speculate that changes in components of the extra- Chang SC, Mulloy B, Magee AI, Couchman JR (2011) Two distinct sites in cellular matrix might represent an important level of regulation sonic Hedgehog combine for heparan sulfate interactions and cell signal- of Shh activity. In support of this, the HSPG biosynthetic enzyme ing functions. J Biol Chem 286:44391–44402. CrossRef Medline Touahri et al. • Sulf1, An Activator of Shh-Dependent OPC Fate Choice J. Neurosci., December 12, 2012 • 32(50):18018–18034 • 18033

Danesin C, Agius E, Escalas N, Ai X, Emerson C, Cochard P, Soula C (2006) Kessaris N, Pringle N, Richardson WD (2001) Ventral neurogenesis and the Ventral neural progenitors switch toward an oligodendroglial fate in re- neuron-glial switch. Neuron 31:677–680. CrossRef Medline sponse to increased Sonic hedgehog (Shh) activity: involvement of Sulfa- Kessaris N, Jamen F, Rubin LL, Richardson WD (2004) Cooperation be- tase 1 in modulating Shh signaling in the ventral spinal cord. J Neurosci tween sonic hedgehog and fibroblast growth factor/MAPK signalling 26:5037–5048. CrossRef Medline pathways in neocortical precursors. Development 131:1289–1298. Das RM, Van Hateren NJ, Howell GR, Farrell ER, Bangs FK, Porteous VC, CrossRef Medline Manning EM, McGrew MJ, Ohyama K, Sacco MA, Halley PA, Sang HM, Kessaris N, Fogarty M, Iannarelli P, Grist M, Wegner M, Richardson WD Storey KG, Placzek M, Tickle C, Nair VK, Wilson SA (2006) A robust (2006) Competing waves of oligodendrocytes in the forebrain and post- system for RNA interference in the chicken using a modified microRNA natal elimination of an embryonic lineage. Nat Neurosci 9:173–179. operon. Dev Biol 294:554–563. CrossRef Medline CrossRef Medline Deneen B, Ho R, Lukaszewicz A, Hochstim CJ, Gronostajski RM, Anderson Kessaris N, Pringle N, Richardson WD (2008) Specification of CNS glia DJ (2006) The transcription factor NFIA controls the onset of gliogen- from neural stem cells in the embryonic neuroepithelium. Philos Trans R esis in the developing spinal cord. Neuron 52:953–968. CrossRef Medline Soc Lond B Biol Sci 363:71–85. CrossRef Medline Dessaud E, Yang LL, Hill K, Cox B, Ulloa F, Ribeiro A, Mynett A, Novitch BG, Kuhlbrodt K, Herbarth B, Sock E, Hermans-Borgmeyer I, Wegner M (1998) Briscoe J (2007) Interpretation of the sonic hedgehog morphogen gra- Sox10, a novel transcriptional modulator in glial cells. J Neurosci 18:237– dient by a temporal adaptation mechanism. Nature 450:717–720. 250. Medline CrossRef Medline Lamanna WC, Baldwin RJ, Padva M, Kalus I, Ten Dam G, van Kuppevelt TH, Dhoot GK, Gustafsson MK, Ai X, Sun W, Standiford DM, Emerson CP Jr Gallagher JT, von Figura K, Dierks T, Merry CL (2006) Heparan sulfate (2001) Regulation of Wnt signaling and embryo patterning by an extra- 6-O-endosulfatases: discrete in vivo activities and functional co- cellular sulfatase. Science 293:1663–1666. CrossRef Medline operativity. Biochem J 400:63–73. CrossRef Medline Dierker T, Dreier R, Migone M, Hamer S, Grobe K (2009) Heparan sulfate Lamanna WC, Kalus I, Padva M, Baldwin RJ, Merry CL, Dierks T (2007) and transglutaminase activity are required for the formation of covalently The heparanome—the enigma of encoding and decoding heparan sulfate cross-linked hedgehog oligomers. J Biol Chem 284:32562–32571. sulfation. J Biotechnol 129:290–307. CrossRef Medline CrossRef Medline Langseth AJ, Munji RN, Choe Y, Huynh T, Pozniak CD, Pleasure SJ (2010) Dubois L, Bally-Cuif L, Crozatier M, Moreau J, Paquereau L, Vincent A Wnts influence the timing and efficiency of oligodendrocyte precursor (1998) XCoe2, a transcription factor of the Col/Olf-1/EBF family in- cell generation in the telencephalon. J Neurosci 30:13367–13372. volved in the specification of primary neurons in Xenopus. Curr Biol CrossRef Medline 8:199–209. CrossRef Medline Lei Q, Jeong Y, Misra K, Li S, Zelman AK, Epstein DJ, Matise MP (2006) Ericson J, Thor S, Edlund T, Jessell TM, Yamada T (1992) Early stages of Wnt signaling inhibitors regulate the transcriptional response to mor- motor neuron differentiation revealed by expression of homeobox gene phogenetic Shh-Gli signaling in the neural tube. Dev Cell 11:325–337. Islet-1. Science 256:1555–1560. CrossRef Medline CrossRef Medline Ericson J, Morton S, Kawakami A, Roelink H, Jessell TM (1996) Two critical Lek M, Dias JM, Marklund U, Uhde CW, Kurdija S, Lei Q, Sussel L, Ruben- periods of Sonic Hedgehog signaling required for the specification of stein JL, Matise MP, Arnold HH, Jessell TM, Ericson J (2010) A home- motor neuron identity. Cell 87:661–673. CrossRef Medline odomain feedback circuit underlies step-function interpretation of a Shh Farshi P, Ohlig S, Pickhinke U, Ho¨ing S, Jochmann K, Lawrence R, Dreier R, morphogen gradient during ventral neural patterning. Development 137: Dierker T, Grobe K (2011) Dual roles of the Cardin-Weintraub motif in 4051–4060. CrossRef Medline multimeric Sonic hedgehog. J Biol Chem 286:23608–23619. CrossRef Li H, de Faria JP, Andrew P, Nitarska J, Richardson WD (2011) Phosphor- Medline ylation regulates OLIG2 cofactor choice and the motor neuron- Freeman SD, Moore WM, Guiral EC, Holme AD, Turnbull JE, Pownall ME oligodendrocyte fate switch. Neuron 69:918–929. CrossRef Medline (2008) Extracellular regulation of developmental cell signaling by Xt- Liu Z, Hu X, Cai J, Liu B, Peng X, Wegner M, Qiu M (2007) Induction of Sulf1. Dev Biol 320:436–445. CrossRef Medline oligodendrocyte differentiation by Olig2 and Sox10: evidence for recip- Fu H, Qi Y, Tan M, Cai J, Takebayashi H, Nakafuku M, Richardson W, Qiu M rocal interactions and dosage-dependent mechanisms. Dev Biol 302:683– (2002) Dual origin of spinal oligodendrocyte progenitors and evidence 693. CrossRef Medline for the cooperative role of Olig2 and Nkx2.2 in the control of oligoden- Lu QR, Yuk D, Alberta JA, Zhu Z, Pawlitzky I, Chan J, McMahon AP, Stiles drocyte differentiation. Development 129:681–693. Medline CD, Rowitch DH (2000) Sonic hedgehog-regulated oligodendrocyte Garel S, Marín F, Matte´i MG, Vesque C, Vincent A, Charnay P (1997) Fam- lineage genes encoding bHLH proteins in the mammalian central nervous ily of Ebf/Olf-1-related genes potentially involved in neuronal differenti- system. Neuron 25:317–329. CrossRef Medline ation and regional specification in the central nervous system. Dev Dyn Mekki-Dauriac S, Agius E, Kan P, Cochard P (2002) Bone morphogenetic 210:191–205. CrossRef Medline proteins negatively control oligodendrocyte precursor specification in the Gill R, Hitchins L, Fletcher F, Dhoot GK (2010) Sulf1A and HGF regulate chick spinal cord. Development 129:5117–5130. Medline satellite-cell growth. J Cell Sci 123:1873–1883. CrossRef Medline Miller RH, Dinsio K, Wang R, Geertman R, Maier CE, Hall AK (2004) Pat- Goodrich LV, Johnson RL, Milenkovic L, McMahon JA, Scott MP (1996) terning of spinal cord oligodendrocyte development by dorsally derived Conservation of the hedgehog/patched signaling pathway from flies to BMP4. J Neurosci Res 76:9–19. CrossRef Medline mice: induction of a mouse patched gene by Hedgehog. Genes Dev 10: Morimoto-Tomita M, Uchimura K, Werb Z, Hemmerich S, Rosen SD 301–312. CrossRef Medline (2002) Cloning and characterization of two extracellular heparin- Gritli-Linde A, Lewis P, McMahon AP, Linde A (2001) The whereabouts of degrading endosulfatases in mice and humans. J Biol Chem 277: a morphogen: direct evidence for short- and graded long-range activity of 49175–49185. CrossRef Medline hedgehog signaling peptides. Dev Biol 236:364–386. CrossRef Medline Nagamine S, Tamba M, Ishimine H, Araki K, Shiomi K, Okada T, Ohto T, Hamburger V, Hamilton HL (1992) A series of normal stages in the devel- Kunita S, Takahashi S, Wismans RG, van Kuppevelt TH, Masu M, Keino- opment of the chick embryo. 1951. Dev Dyn 195:231–272. CrossRef Masu K (2012) Organ-specific sulfation patterns of heparan sulfate gen- Medline erated by extracellular sulfatases Sulf1 and Sulf2 in mice. J Biol Chem Hochstim C, Deneen B, Lukaszewicz A, Zhou Q, Anderson DJ (2008) Iden- 287:9579–9590. CrossRef Medline tification of positionally distinct astrocyte subtypes whose identities are Novitch BG, Chen AI, Jessell TM (2001) Coordinate regulation of motor specified by a homeodomain code. Cell 133:510–522. CrossRef Medline neuron subtype identity and pan-neuronal properties by the bHLH re- Kalus I, Salmen B, Viebahn C, von Figura K, Schmitz D, D’Hooge R, Dierks T pressor Olig2. Neuron 31:773–789. CrossRef Medline (2009) Differential involvement of the extracellular 6-O-endosulfatases Orentas DM, Hayes JE, Dyer KL, Miller RH (1999) Sonic hedgehog signal- Sulf1 and Sulf2 in brain development and neuronal and behavioural plas- ing is required during the appearance of spinal cord oligodendrocyte ticity. J Cell Mol Med 13:4505–4521. CrossRef Medline precursors. Development 126:2419–2429. Medline Kang P, Lee HK, Glasgow SM, Finley M, Donti T, Gaber ZB, Graham BH, Park HC, Shin J, Appel B (2004) Spatial and temporal regulation of ventral Foster AE, Novitch BG, Gronostajski RM, Deneen B (2012) Sox9 and spinal cord precursor specification by Hedgehog signaling. Development NFIA coordinate a transcriptional regulatory cascade during the initia- 131:5959–5969. CrossRef Medline tion of gliogenesis. Neuron 74:79–94. CrossRef Medline Poncet C, Soula C, Trousse F, Kan P, Hirsinger E, Pourquie´ O, Duprat AM, 18034 • J. Neurosci., December 12, 2012 • 32(50):18018–18034 Touahri et al. • Sulf1, An Activator of Shh-Dependent OPC Fate Choice

Cochard P (1996) Induction of oligodendrocyte progenitors in the Stolt CC, Lommes P, Sock E, Chaboissier MC, Schedl A, Wegner M (2003) trunk neural tube by ventralizing signals: effects of notochord and floor The Sox9 transcription factor determines glial fate choice in the develop- plate grafts, and of sonic hedgehog. Mech Dev 60:13–32. CrossRef ing spinal cord. Genes Dev 17:1677–1689. CrossRef Medline Medline Stolt CC, Lommes P, Friedrich RP, Wegner M (2004) Transcription fac- Pringle NP, Yu WP, Guthrie S, Roelink H, Lumsden A, Peterson AC, Rich- tors Sox8 and Sox10 perform non-equivalent roles during oligoden- ardson WD (1996) Determination of neuroepithelial cell fate: induction drocyte development despite functional redundancy. Development of the oligodendrocyte lineage by ventral midline cells and sonic hedge- 131:2349–2358. CrossRef Medline hog. Dev Biol 177:30–42. CrossRef Medline Sun T, Echelard Y, Lu R, Yuk DI, Kaing S, Stiles CD, Rowitch DH (2001) Olig Qi Y, Cai J, Wu Y, Wu R, Lee J, Fu H, Rao M, Sussel L, Rubenstein J, Qiu M bHLH proteins interact with the homeodomain proteins to regulate cell (2001) Control of oligodendrocyte differentiation by the Nkx2.2 home- fate acquisition in progenitors of the ventral neural tube. Curr Biol 11: odomain transcription factor. Development 128:2723–2733. Medline 1413–1420. CrossRef Medline Qi Y, Tan M, Hui CC, Qiu M (2003) Gli2 is required for normal Shh signal- Takebayashi H, Yoshida S, Sugimori M, Kosako H, Kominami R, Nakafuku ing and oligodendrocyte development in the spinal cord. Mol Cell Neu- M, Nabeshima Y (2000) Dynamic expression of basic helix-loop-helix rosci 23:440–450. CrossRef Medline Ratzka A, Kalus I, Moser M, Dierks T, Mundlos S, Vortkamp A (2008) Re- Olig family members: implication of Olig2 in neuron and oligodendro- dundant function of the heparan sulfate 6-O-endosulfatases Sulf1 and cyte differentiation and identification of a new member, Olig3. Mech Dev Sulf2 during skeletal development. Dev Dyn 237:339–353. CrossRef 99:143–148. CrossRef Medline Medline Tanabe Y, William C, Jessell TM (1998) Specification of motor neuron Richardson WD, Smith HK, Sun T, Pringle NP, Hall A, Woodruff R (2000) identity by the MNR2 homeodomain protein. Cell 95:67–80. CrossRef Oligodendrocyte lineage and the motor neuron connection. Glia 29:136– Medline 142. CrossRef Medline Vallstedt A, Klos JM, Ericson J (2005) Multiple dorsoventral origins of oli- Rowitch DH (2004) Glial specification in the vertebrate neural tube. Nat godendrocyte generation in the spinal cord and hindbrain. Neuron 45: Rev Neurosci 5:409–419. CrossRef Medline 55–67. CrossRef Medline Rowitch DH, Kriegstein AR (2010) Developmental genetics of vertebrate Viviano BL, Paine-Saunders S, Gasiunas N, Gallagher J, Saunders S (2004) glial-cell specification. Nature 468:214–222. CrossRef Medline Domain-specific modification of heparan sulfate by Qsulf1 modulates the Rubin JB, Choi Y, Segal RA (2002) Cerebellar proteoglycans regulate sonic binding of the bone morphogenetic protein antagonist Noggin. J Biol hedgehog responses during development. Development 129:2223–2232. Chem 279:5604–5611. CrossRef Medline Medline Vokes SA, Ji H, McCuine S, Tenzen T, Giles S, Zhong S, Longabaugh WJ, Scardigli R, Schuurmans C, Gradwohl G, Guillemot F (2001) Crossregula- Davidson EH, Wong WH, McMahon AP (2007) Genomic characterization between Neurogenin2 and pathways specifying neuronal identity in tion of Gli-activator targets in sonic hedgehog-mediated neural pattern- the spinal cord. Neuron 31:203–217. CrossRef Medline ing. Development 134:1977–1989. CrossRef Medline Scott CE, Wynn SL, Sesay A, Cruz C, Cheung M, Gomez Gaviro MV, Booth S, Wojcinski A, Nakato H, Soula C, Glise B (2011) DSulfatase-1 fine-tunes Gao B, Cheah KS, Lovell-Badge R, Briscoe J (2010) SOX9 induces and Hedgehog patterning activity through a novel regulatory feedback loop. maintains neural stem cells. Nat Neurosci 13:1181–1189. CrossRef Dev Biol 358:168–180. CrossRef Medline Medline Yan D, Lin X (2009) Shaping morphogen gradients by proteoglycans. Cold Soula C, Danesin C, Kan P, Grob M, Poncet C, Cochard P (2001) Distinct Spring Harb Perspect Biol 1:a002493. CrossRef Medline sites of origin of oligodendrocytes and somatic motoneurons in the chick spinal cord: oligodendrocytes arise from Nkx2.2-expressing progenitors Zhang F, McLellan JS, Ayala AM, Leahy DJ, Linhardt RJ (2007) Kinetic and by a Shh-dependent mechanism. Development 128:1369–1379. Medline structural studies on interactions between heparin or heparan sulfate and St-Jacques B, Dassule HR, Karavanova I, Botchkarev VA, Li J, Danielian PS, proteins of the hedgehog signaling pathway. Biochemistry 46:3933–3941. McMahon JA, Lewis PM, Paus R, McMahon AP (1998) Sonic hedgehog CrossRef Medline signaling is essential for hair development. Curr Biol 8:1058–1068. Zhou Q, Wang S, Anderson DJ (2000) Identification of a novel family of CrossRef Medline oligodendrocyte lineage-specific basic helix-loop-helix transcription fac- Stolt CC, Rehberg S, Ader M, Lommes P, Riethmacher D, Schachner M, tors. Neuron 25:331–343. CrossRef Medline Bartsch U, Wegner M (2002) Terminal differentiation of myelin- Zhou Q, Choi G, Anderson DJ (2001) The bHLH transcription factor Olig2 forming oligodendrocytes depends on the transcription factor Sox10. promotes oligodendrocyte differentiation in collaboration with Nkx2.2. Genes Dev 16:165–170. CrossRef Medline Neuron 31:791–807. CrossRef Medline